| Commit message (Collapse) | Author | Age |
| |
|
|
|
|
|
|
|
|
|
| |
addressing:
i. some naming conventions
ii. add macro to guard cpp files
TODO:
cpplint
|
|
|
|
| |
followup commit.
|
|
|
|
|
|
|
|
| |
addressing code review comments.
TODO:
MACRO GOOGLE_TENSORRT to guard c++ files.
build/test -> current code works on local repo before the master merge.
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
* Decoding contents of BMP file on big endian
* Updated as per review comments
* Update decode_bmp_op.cc
Corrected function name
|
| |
| |
| |
| |
| |
| | |
* Update contrib/HVX readme
* Fix PR comments
|
| | |
|
| |\
| | |
| | | |
Branch 183374082
|
| |/| |
|
| | |
| | |
| | |
| | |
| | |
| | | |
intrinsics.
PiperOrigin-RevId: 183374082
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183374040
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183367326
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183354193
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183347199
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183345779
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183342483
|
| | |
| | |
| | |
| | |
| | |
| | | |
build failures.
PiperOrigin-RevId: 183341561
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183339087
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
* fixed #15487
* fixed #15487
* simplify and tweak formatting to TF style
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
On line 83 we test to see if single_value is an int (or able to be cast to an int).
ValueError is fired if `int()` is called with an input like 'asdf' - this is caught and gives a helpful error, using the 'name' param to provide more context.
However, when given an other than a string or int, this is *not* caught - making error messages much more esoteric than the helpful one written out here.
For example, before, I was getting an error:
> line 83, in normalize_tuple
> int(single_value)
> TypeError: int() argument must be a string or a number, not 'tuple'
Now I get the more useful:
> ValueError: The `kernel_size` argument must be a tuple of 2 integers.
> Received: ((0, 3), 50) including element (0, 3) of type <type 'tuple'>
|
| | | |
|
| | | |
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183333452
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183333411
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183331075
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183328456
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183328052
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
* Including common.h with NEON_2_SSE.h
Including common.h to make sure that USE_NEON is defined in case of NEON_2_SSE.h is used; otherwise USE_NEON will not be propagated to this file and `portable_tensor_utils.h` will be used
* Removing spaces
* Using neon is USE_NEON is defined
* Using common.h to set USE_NEON if applicable
* Adding required dependencies
Since tests are more likely compiled on sse4-enabled machine - if they are run with common.h - it will enable sse4, there are two ways - just enable it and let tests using sse4, or add extra flag check which would require USE_SSE4 flag to be defined. I've added all required dependencies for tests to pass, so it could work.
* Rearranging dependencies according to linter
* Fixing sanity check
* trying to rerun, could be cached
* Triggering build
* triggering build - looks like it stuck
no status for more than 24h
* including x86_64 since NEON_2_SSE4 convertion will allow neon to run on x86 with SSE3,4
* Fixing name
* using darwing for MacOS
* Fixing header inclusion rules
* Fix syntax
* Depend on includes
We depend on both types and compatibility. That's in the "types" rule.
|
| | |
| | |
| | |
| | |
| | |
| | | |
* fix typos
* fix typos
|
| | | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
//third_party/tensorflow/python/data/kernel_tests:dataset_from_generator_op_test to medium
It sometimes takes longer than a minute, and thus gets flaky timeouts.
PiperOrigin-RevId: 183326334
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
unlike the existing --default_ranges_{min,max} flags which only allowed
to set a single global value for all arrays.
This takes the form of a new embedded message in ModelFlags, which is
its own message so that it can be serialized separately. The command-line
interface is --arrays_extra_info_file=some_proto.pbtxt, i.e. we don't
try to make a command-line-flags-only interface, we mandate putting the info
in a file. The rationale is that users may want to specify custom minmax
for hundreds of arrays, so it would be cumbersome to have that all in a
command line.
This should be considered an experimental feature, in the sense that
in properly quantized models, minmax information is already embedded
in the graph (e.g. in FakeQuant nodes). This is an extension of the
existing --default_ranges_{min,max} feature which had turned out to be
too restrictive for many users.
PiperOrigin-RevId: 183326000
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183321394
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
initializing variables.
Today, when variables are constructed (Resource and otherwise), we lift certain
operations, including the VarHandleOp and initialization ops, out of all
control flow contexts; the mechanism for doing so is entering the context
manager returned by `control_dependencies(None)`. This change replaces various
instances of this mechanism with `init_scope`, which clears control dependencies, lifts ops out of function-building graphs, and pauses the gradient tape. As a result,
variables that are created inside graph functions will be automatically hoisted
into an outer context.
PiperOrigin-RevId: 183320576
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183318384
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183317862
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This also moves the set_producer_version function from a specific test
file to test_util.py, since it's needed in two test files now.
PiperOrigin-RevId: 183316990
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183315762
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183313321
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183312680
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183312596
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
* import tensorflow as tf
* import tensorflow as tf
* from contextlib import contextmanager
* remove the last remaining change to py2tf
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This fixes the original rollback by using placeholders for the
SparseTensor shapes. The flakiness was caused by the nondeterministic
ordering of the sequences dict.
PiperOrigin-RevId: 183308774
|
| | |
| | |
| | |
| | |
| | |
| | | |
Without it we trigger a segmentation fault, but later in a different stack, which is not so helpful.
PiperOrigin-RevId: 183307729
|
| | |
| | |
| | |
| | |
| | |
| | | |
match the number of outputs.
PiperOrigin-RevId: 183306267
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
* Implement LoggingAsync for GRPC Worker.
* Add nullptr checks
* Fix BUILD file format
* Change LoggingAsync implementation
- Revert changes to *_rendezvous_mgr
- Implement logging primitives in session_mgr instead
- Implement ClearLogs
- Fixed C++ formating
* Check for nullptr
* Better handling of the case when both "clear" and "retrieve" flags are sent to "LogAsync."
nullptr check on default_worker_cache_.
* Fix formatting
* Updata session_mgr.cc to address changes in 619792f
|
| | | |
|
| | |
| | |
| | | |
The local Dockerfile does not have all the dependencies for running the exercise notebooks in udacity assignments.
|
| | |
| | |
| | |
| | | |
PiperOrigin-RevId: 183301479
|